Research On Transmission Optimization Technology In Delay Tolerant Networks

With the development of communication technologies, communication networks in challenged environments increasingly emerge, such as underwater network communications, wireless sensor network communications, deep space network communications and so on. Data communications under such environments typically suffer from large propagation delay, high bit error rate, low signal/carrier to noise ratio and frequent link disruptions. The network transmission performance is facing enormous challenges. Affected by environmental factors, the traditional routing and transport mechanisms in this environment are inefficient or even cannot work properly.Recently, Delay Tolerant Network (DTN) has been proposed to overcome challenges related to efficient and reliable data delivery in wireless networks in challenged environments. Due to the diverse and complex environmental factors, the transmission performance of DTN faces more challenges compared to traditional wireless networks. Firstly, nodes movement and the occlusion of environmental objects incur the absence of the end-to-end communication path. The link between communication nodes interrupts frequently. The node must take full use of the link connectivity opportunities to accomplish data delivery. Secondly, affected by the long communication distance and environmental noise, the information data transmission rate between nodes is low and the bandwidth is limited. To improve the data delivery efficiency, it is necessary to improve the utilization of the limited bandwith resources. Thirdly, affected by the weight and the cost of communication nodes, the capacity and storage resources of DTN nodes are limited. However, the store and forward mechanism of the Bundle protocol requires large node resources. In order to ensure the successful delivery of data among nodes, a proper management of DTN node resources is required.This paper focuses on the problem of low goodput and low storage efficiency when data trasmitting in DTN in extreme environments. We propose a generic model for the analysis of the transmission performance of DTN, in which we model the mechanism of the LTP protocol, the store and forward characteristic of the Bundle protocol and the release process of the node storage. Based on that, we propose two mechanisms, which are the end-to-end goodput enhancement mechanism and the node storage management mechanism. These two mechanisms are analyzed and simulated under deep space channel conditions. The analytic model proposed in this paper can be widely applied to the analysis and research of the transmission performance of DTN based on Bundle protocol and LTP protocol. The main contributions of this paper are listed as follows.(1) The goodput when the application data delivering in the multi-hop path is studied in this paper. A novel end-to-end goodput enhancement mechanism is proposed. The mechanism can maximize the end-to-end goodput by adaptively adjusting Bundle size and LTP frame size at each hop according to the link state. Theoretical analysis shows that the end-to-end goodput is not only affected by the link state, but also the size of protocol data unit at various network layers. We obtain the Bundle single-hop delivery delay model and the application data multi-hop goodput model by theoretically modeling the process of Bundle delivering between nodes and application data delivering among nodes. Based on that, we proposed the end-to-end goodput enhancement mechanism. Simulation results under deep space channel conditions show that our mechanism can effectively improve the end-to-end goodput when application data delivering in the multi-hop path compared to traditional schemes where joint packet size optimization is not considered.(2) The release rate of node storage resources when data delivering between two DTN nodes is also studied in this paper. A novel node storage management mechanism is proposed. The mechanism can effectively improve the release rate of node storage resources by adaptively adjusting Bundle size and LTP frame size according to the link state between two DTN nodes. Theoretical analysis shows that the release rate of node storage resources is affected by the link state, the size of protocol data unit at various network layers, the peaking storage during data delivering and the node capacity. We obtain the node storage release model by theoretically modeling the releasing process of node storage resources when data delivering between nodes. Based on that, we proposed the node storage management mechanism. Simulation results under deep space channel conditions show that our mechanism can effectively improve the release rate of node storage resources compared to traditional schemes where joint packet size optimization is not considered, which means our mechanism is suitable to extreme DTN environments.